Maria Lidqvist

Maria Lidqvist

Department of Microbiology and Immunology Institute of Biomedicine

Sahlgrenska Academy at University of Gothenburg

Gothenburg 2012

Monoclonal antibodies against human papillomavirus E7 oncoprotein for diagnosis of cervical neoplasia and cancer

© Maria Lidqvist 2012 maria.lidqvist@fdab.com ISBN 978-91-628-8511-3

Printed in Gothenburg, Sweden 2012 Ineko AB, Gothenburg

For my family

Maria Lidqvist

Department of Microbiology and Immunology, Institute of Biomedicine Sahlgrenska Academy at University of Gothenburg, Göteborg, Sweden

Abstract

Cervical cancer is the second most common cancer among women worldwide with half a million of new cases every year. Cervical cancer is caused by oncogenic human papillomaviruses (HPVs), with HPV16 and 18 being the most frequently detected types. Genital HPV infections are common in the general population although most infections are cleared before causing malignancy. However a small proportion of the infections evade the immune system, become persistent and may cause cervical lesions and even invasive disease. Being the causative agent for cervical malignancy, HPV is an obvious target for cervical cancer diagnosis and prevention. Current screening programs, primarily based on cervical cytology, produce millions of suspicious samples every year. Specific tools to identify high- grade disease in these samples are needed to increase specificity for malignancy and thereby reduce referral rates and overtreatment.

In the current study, monoclonal antibodies were raised against the HPV E7 oncoprotein. E7 is an absolute prerequisite for malignant transformation and the protein is expressed at increasing levels during cancer development. E7 is therefore a suitable marker for HPV-induced malignancy. Antibodies specific for the E7 protein of oncogenic HPV types were selected using immunological methods such as ELISA, Western blot, Immunocytochemistry and flow cytometry. Phage display was used to identify antibody epitopes thereby predicting and verifying antibody specificity. Two of the antibodies, recognizing HPV16 and 18 E7 respectively, demonstrated strong staining of dysplastic cells in HPV-positive specimens in immunocytochemistry and may thus have the potential to be used in a clinical setting. Since the antibodies detect the protein in Liquid-based cytology, which normally leaves residual sample after standard cytology, E7 testing can easily be performed without recalling the patient for additional sampling.

Immunological detection of the E7 oncoprotein is an attractive alternative for triage of suspicious and borderline cytology to highlight and identify the often rare dysplastic cells present in a cell scrape. E7 detection can further reduce subjectivity and be performed with only standard equipment and thereby make HPV-testing available also in less developed regions.

Keywords: Human papillomavirus (HPV), Cervical cancer, Cervical neoplasia, E7 oncoprotein, Monoclonal antibodies, Cytology, Immunocytochemistry, Phage display ISBN: 978-91-628-8511-3

http://hdl.handle.net/2077/29710

I. Maria Lidqvist, Olle Nilsson, Jan Holmgren, Christina Hall, Christian Fermér. Phage display for site-specific immunization and characterization of high-risk human papillomavirus specific E7 monoclonal antibodies.

Journal of Immunological Methods, 337 (2008) 88-96.

II. Maria Lidqvist, Olle Nilsson, Jan Holmgren, Sebastian Hölters, Eva Röijer, Matthias Dürst, Christian Fermér. Detection of human papillomavirus oncoprotein E7 in liquid-based cytology.

Journal of General Virology, 93 (2012) 356–363.

III. Michael Lebens, Susanne Källgård, Christian Fermer,

Maria Lidqvist, Hubert Bernauer. Generation of plasmid encoded protein-specific overlapping peptide libraries and the mapping of B cell epitopes in HPV E7 oncoprotein recognized by monoclonal antibodies and antibodies in the serum of immunized mice.

Submitted manuscript.

IV. Maria Lidqvist, Olle Nilsson, Jan Holmgren, Matthias Dürst, Elin Andersson, Peter Horal, Christian Fermér. Clinical feasibility study of anti-E7 MAb for detection of HPV-induced neoplasia and cancer in liquid-based cytology. Submitted manuscript.

Reprints were made with permissions from the publishers.

Ab Antibody

ASC-US Atypical Squamous Cells of Undetermined Significance ATCC American Type Culture Collection

CIN Cervical Intraepithelial Neoplasia CR Conserved region

CTB Cholera toxin B subunit

E Early

GST Glutathione-S-Transferase HPV Human papillomavirus HRP Horseradish peroxidase ICC Immunocytochemistry

Ig Immunoglobulin

L Late

LCR Long control region mAb Monoclonal antibody ORF Open reading-frame PEG Polyethylene glycol pRb Retinoblastoma protein SCC Squamous cell carcinoma

1.1 Human papillomaviruses & Cervical cancer ... 2

1.1.1 The Human papillomavirus ... 2

1.1.2 Natural history of HPV infections ... 3

1.1.3 From HPV infection to cancer ... 5

1.1.4 HPV E6 and E7 oncoproteins ... 8

1.2 Cervical cancer prevention ... 10

1.2.1 Screening and diagnosis ... 10

1.2.2 Cervical cancer screening in a vaccinated population ... 10

1.2.3 Challenges in cervical cancer prevention ... 11

1.2.4 Biomarkers in cervical screening ... 12

2 AIM ... 14

3 KEY METHODOLOGIES ... 15

3.1 Cervical cancer cell lines ... 15

3.2 Generation, identification and production of monoclonal antibodies . 15 3.3 Phage display ... 17

3.3.1 Site-specific immunization ... 17

3.3.2 Epitope determination using overlapping E7 peptides ... 17

3.3.3 Mimotope determination using random peptide libraries ... 18

3.4 Construction of a vector system for the display of peptide libraries ... 18

3.5 ELISA ... 20

3.6 Western blot ... 21

3.7 Immunocytochemistry ... 21

3.8 Staining of clinical LBC samples ... 21

4 RESULTS &DISCUSSION ... 23

4.1 Site-specific immunization ... 25

4.2 Immunization with full-length E7 and the identification of high-risk specific E7 antibodies ... 28

4.4 Detection of E7 in LBC samples ... 32

4.4.1 Protocol optimization ... 32

4.4.2 Evaluation on clinical LBC samples ... 34

5 CONCLUDING REMARKS ... 36

ACKNOWLEDGEMENTS ... 39

REFERENCES ... 41

Cervical cancer is the second most common cancer in women worldwide, with half a million new cases every year. Cervical cancer is caused by a limited number of oncogenic human papillomaviruses (HPVs) and viral DNA can be detected in almost all cancers (99.7 %)¹. Most women and men are infected with genital HPV at least once in their life-time and the prevalence of genital HPV is over 10 % in the world². The vast majority of all infections are transient and/or asymptomatic. However, a small proportion will evade the immune system and if left untreated, progress to cancer. The major challenge in cervical cancer diagnosis and management is thus to discriminate persistent high-risk infections, that should be treated or closely monitored, from transient infections and low-grade lesions that will regress, while minimizing overtreatment and unnecessary recalling of patients.

The Pap test is the most widely used screening method, often complemented by colposcopy and histology. In the Pap test, the morphology of exfoliated cervical cells is studied to identify pre-malignant and cancerous cells. Pap screening has reduced the number of cancer cases dramatically in many areas^{3, 4}, but to be efficient the test has to be repeated frequently and requires trained personnel. Millions of pap tests yield inconclusive or borderline results every year and require further sampling and evaluation to identify underlying disease. In addition, many women in the world are not included in a cervical cancer prevention program and thus cervical cancer still causes 270 000 deaths every year⁵.

To reduce the number of cancer cases, both in regions with and without functional screening, more cost-effective, cancer-specific and preferably objective diagnostic methods are needed. Since Harald zur Hausen’s discovery of the causality between high-risk HPV infection and cervical cancer in 1980, for which he was awarded with the Nobel Prize in physiology or medicine in 2008, a lot of effort has been put into understanding the viral life cycle and its ability to induce malignancy. Based on this knowledge HPV has become an important target for diagnosis and prevention of cervical cancer.

This thesis describes the generation of monoclonal antibodies (mAbs) detecting the E7 oncoprotein of carcinogenic HPV types and the characterization of the antibodies with special reference to their ability to specifically recognize E7 of high-risk types and finally a preliminary evaluation of their usefulness for immunological detection of E7 oncoprotein

in clinical specimens. Different immunization and selection strategies were used to obtain antibodies with desired functionalities. Antibodies were identified that proved to be useful in different immunological methods for E7 detection and can hopefully be used for development of more specific tests for early diagnosis of HPV-induced malignancy.

Papillomaviruses are small, non-enveloped DNA viruses infecting epithelial surfaces in most vertebrates. The viruses cause a broad range of disease in humans, such as anogenital cancer, head- and neck cancer, skin or genital warts and recurrent respiratory papillomatosis. The outcome of the infection will depend on HPV type and site of infection as well as different host factors. HPV is necessary for the development of cervical cancer and viral DNA can be found in virtually all cases¹. The two most common types in cervical cancer are HPV16 and 18, together responsible for approximately 70 % of all cases⁶.

So far, at least 120 HPV types with completely characterized genomes have been identified. The virus is classified based on the nucleotide sequence of the HPV L1 capsid gene and viruses with at least 10 % difference in L1 sequence are defined as different types⁷. The HPV genome evolves slowly, in parallel with its host and has more and less conserved regions such as L1 being well conserved and the noncoding Long control region (LCR) being more diverse⁸.

Different HPV types target different epithelia and at least 40 types infect the human genital tract. Of these, 12 are frequently found in cervical cancers and therefore considered high-risk types⁹. Women, and men, are often infected with several HPV types at the same time, often transmitted together and different infections can cause independent lesions side by side in the cervix^{10, 11}. HPV16 is by far the most dominant HPV type, responsible for 55 % of the cervical cancers in the world, followed by HPV18 present in at least 15 % of the invasive cancers⁶. The other high-risk types are HPV33, 45, 31, 58, 52, 35, 59, 51, 56 and 39 in descending order in the world-wide distribution though there are geographical variations^{6, 12}. The other mucosal types, besides the 12 high-risk types, are classified as intermediate or low- risk and only rarely detected in cervical cancer. HPV6 and 11 are the two most common genital low-risk types together responsible for 90 % of genital warts¹³.

The viruses are prone to cause cancer at transformation zones where two types of epithelia meet, especially the cervix where squamous epithelium replaces glandular¹³. Most cervical cancers are of squamous origin but approximately 15 % are adenocarcinomas. HPV18 is more common in adenocarcinoma than in Squamous cell carcinoma (SCC), detectable in 37 % of cases, compared to 13 % of the SCC⁶. Conversely, HPV16 is less frequently found in adenocarcinoma than in SCC. HPV16, 18 and 45 are often considered more aggressive since they are present in a larger proportion of SCC than in high-grade lesions compared to other HPV types⁶ (Fig. 1).

Invasive cancer caused by HPV16, 18 and 45 is in addition diagnosed at a younger mean age¹².

Figure 1. Type-specific HPV prevalence in relation to cervical diagnoses. HPV16, 18 and 45 are often considered more aggressive than the other high-risk types as the prevalence of these types increase more with the severity of the disease.HPV16, 18 and 45 are further responsible for a larger portion of the adenocarcinomas (ADC) than squamous cell carcinoma (SCC). Multiple infections are counted several times.

HSIL: high-grade squamous intraepithelial lesions (corresponding to Cervical intraepithelial neoplasia 2/3). Reprinted from¹⁴ Copyright (2008), with permission from Elsevier.

The 8 kb circular HPV DNA genome normally encodes two classes of genes, Early (E) genes, E1, E2, E4, E5, E6 and E7, involved in viral replication and cellular transformation and Late (L) genes encoding the L1 and L2 capsid proteins (Fig. 2). To be able to infect, the viral particles have to reach the proliferating basal cells and probably do so through microabrasions in the epithelia, though the exact mechanisms of viral uptake is not fully understood¹⁵. The life cycle of the virus is then totally dependent on the differentiation program of the keratinocytes and cellular DNA synthesis.

During a productive infection, the expression of viral proteins is tightly regulated in order to produce infectious viral particles and avoid the immune system. The early proteins are expressed soon after infection but at low levels and the viral genome is maintained as stable episomes or plasmids, replicating with the cellular DNA in the basal cells¹⁵. In this initial part of the viral life cycle the HPV E1 and E2 proteins are necessary for replication of the viral DNA¹⁵. The infected basal cells undergo asymmetric mitosis, leaving one daughter cell as an undifferentiated proliferating basal cell acting as a reservoir of infection while the other becomes a differentiating suprabasal cell. The E6 and E7 proteins associate with cell cycle regulators and drive the cells, which would otherwise exit the cell cycle as they differentiate, into S-phase¹⁶ and are thus considered oncoproteins. As the infected cells spread throughout the epithelia, the expression of the E6 and E7 oncoproteins is upregulated and the viral DNA increases in copy number.

The capsid proteins are only expressed in the upper epithelial layers, after viral genome amplification¹⁵ and assemble into viral particles comprised of the viral genome, covered by a capsid consisting of 360 copies of the L1 protein and 12 copies of the L2 protein. Papillomaviruses are non-lytic and shed when the infected cells reach the epithelial surface¹⁷. This infectious cycle takes approximately three weeks, the same time as it takes for the basal cells to differentiate and move up the epithelium.

Figure 2. A schematic presentation of the HPV 16 circular DNA genome.

The Early (E) and Late (L) genes and the Long control region (LCR) are shown.

The majority of the HPV infections are subclinical or cause only mild dysplasia, cleared by the immune system within two years although the mechanisms of HPV clearance are not fully understood. The immune response after HPV infection is normally low and/or delayed, most probably because of the low-level expression of the early HPV proteins, production of the capsid proteins only in the upper epithelial layers far from immune cells in the submucosa and the fact that the infection is non-lytic, thereby not inducing an inflammatory immune response¹⁵.

If a high-risk HPV infection is not cleared by the immune system, it becomes persistent and able to induce malignant transformation. Cervical cancer development normally follows defined stages starting with infection with a carcinogenic HPV type followed by the establishment of viral persistence, the development of pre-cancer and culminating in invasion (Fig. 3). Lesions at all stages until invasive cancer can regress, but the chance of clearance decreases with the severity of the lesion.

Figure 3. From HPV infection to cancer. The majority of the genital HPV infections is asymptomatic and cleared before causing visible lesions.

The majority of all mild dysplasia regress spontaneously within less than a year.

A proportion of the high-risk HPV infections will however become persistent and may, if left untreated, proceed to high-grade lesions and invasive cancer.

The estimated life-time risk of acquiring a genital HPV infection is over 50 %¹⁸ and most women are infected soon after becoming sexually active.

The HPV prevalence therefore peaks in young women and is as high as 40 % in some female populations¹⁹. Due to the long pre-cancerous phase and the fact that most infections do not cause malignancy, cancer incidence in young women is however normally low¹⁹. Since the virus is transmitted by skin-to- skin contact, the risk factors for acquiring HPV are linked to sexual behavior, such as the number of partners and other genital infections²⁰. Approximately half of the carcinogenic HPV infections are resolved within six months of exposure and less than 10 % become persistent and are still detectable after

two years¹⁹. The two most important risk factors for cervical cancer are thus persistent infections with oncogenic HPV and lack of screening¹³. Consequently most cases and deaths from cervical cancer are in low-resource regions, where the possibilities to screen and treat are limited. Other minor risk factors for cervical cancer in HPV-infected women are multiparity²¹, inflammation²² and long-term use of oral contraceptives²³. Smoking is a risk factor for SCC but not adenocarcinoma²⁴.

Viral persistence is necessary for carcinogenesis and as the probability of clearance decreases with time from exposure, the risk of invasion increases²⁵ (Fig. 4). For malignant progression however, other genetic events, induced by the virus, must take place in the host genome¹⁵. As discussed below, the viral oncoproteins interact with different cell cycle regulators, increase cell proliferation and induce genomic instability which leads to an accumulation of somatic mutations during the long pre-cancerous phase. The pre-cancer (Cervical intraepithelial neoplasia, CIN2/3) incidence actually peaks approximately 5-15 years after infection and the invasive cancer decades later. Approximately 1 % of all high-risk infections will finally, if left untreated, cause cervical cancer²⁶.

Figure 4. Risk of HPV persistence and cervical cancer progression. The majority of the HPV infections are solved within two years after infection. However, the longer the infection persists, the larger risk of developing high-grade dysplasia. Most high- grade lesions (Cervical intraepithelial neoplasia, CIN3) develop many years after primary infection. Half of the high-grade lesions will, if left untreated proceed to invasive cancer and true CIN3 should be the preferred threshold for treatment.

Reprinted from¹⁹ Copyright (2010), with permission from Wolters Kluwer Health.

Cervical cancer precursors are often histologically classified as Cervical Intraepithelial Neoplasia (CIN) grade 1-3, where CIN1 is mild, CIN2 moderate and CIN3 severe dysplasia²⁷, defined as the extension of abnormal, undifferentiated cells in the epithelium caused by persistent HPV infections

(Fig. 5). Low-grade lesions (CIN1) are the first manifestation of the HPV infections, caused by both high- and low-risk HPV types. CIN1 should not be considered as pre-cancer and has a relatively low risk of progression to invasive disease. The abnormal cells only occupy the deepest third of the epithelium and these mild to subclinical changes are hard to detect histologically. CIN2 is a heterogeneous state, often with a mix of acute infection and severe dysplasia, with abnormal cells in two thirds of the epithelium. In CIN3, or carcinoma in situ, the undifferentiated cells are spread throughout the epithelia and this is the immediate precursor of invasive cancer and should always be treated due to the high risk of progression.

A key event during carcinogenesis is integration of the viral DNA into the host genome. This is seen in most cancers as well as in many high-grade lesions and some low-grade lesions²⁸. Integration is probably not necessary for malignancy since not all cancers have integrated viral genome, but leads to a more aggressive progression. Integration often leads to disruption of the E2 gene, which otherwise is a regulator of the E6/E7 expression. This loss of negative feedback control of the expression of the oncogenes results in elevated oncoprotein levels¹⁵. Integration cannot be considered a natural event in the viral life cycle since no viral particles are produced from this truncated integrated form.

Figure 5. The expression pattern of the viral proteins changes during the progression to cervical cancer. In the productive infection, viral particles are produced in the upper epithelial layers and shed. The early viral proteins are expressed, at moderate levels, in the lower epithelium. In high-grade dysplasia and cancer, there is no or very low expression of the capsid proteins, while the expression of the oncoproteins is upregulated, especially after viral integration. CIN: Cervical intraepithelial lesion; CaCx: Cervical cancer. Reproduced with permission from John Doorbar (2006) Clinical Science (110) p 533 © the Biochemical Society.

Several studies have shown that two viral proteins, E6 and E7, are necessary both for the induction of the malignant transformation and for maintaining the transformed phenotype^{29, 30}. The E6 and E7 oncoproteins are expressed at increasing levels during cervical carcinogenesis^{31, 32} and are accepted markers for cervical cancer progression. E6 and E7 have numerous cellular targets, with the tumour suppressor p53 and the retinoblastoma protein family (pRb) being the most studied. By interacting with important cell cycle regulators, the virus overrides cell cycle check points and induces cell proliferation. The viral oncoproteins further inhibit apoptosis, induce genomic instability and increase the telomerase expression, important steps in the transformation towards cancer phenotype (reviewed by¹⁶) (Fig. 6).

Figure 6. Illustration of key steps in HPV induced malignancy. In the normal cell, cell cycle progression is controlled by the retinoblastoma protein (pRb) and the E2F transcription factor. The E7 protein can, by degrading the tumour suppressor pRb, release E2F and induce S-phase entry in HPV-infected epithelial cells. Due to loss of feed-back, the levels of tumour suppressor p53 increase causing growth inhibition and apoptosis. When E6 is present however, p53 is degraded which leads to extended proliferation, loss of p53-mediated DNA damage control and accumulation of mutations. E6 can further activate the catalytic subunit of telomerase (hTERT) which is normally inactive in parabasal cells, causing telomere extension and

immortalization¹⁶.

E6 and E7 of high- and low-risk HPVs have large sequence similarities and share some interaction partners (e.g. pRb). However, the low-risk HPV proteins has lower affinity to the cellular proteins and the low- (and intermediate) risk HPV types are not able to induce malignant transformation^{33, 34} and are seldom detected in cervical cancer.

The E7 proteins are small acidic proteins, of approximately 100 amino acids (aa), with more and less conserved regions (Fig. 7). The N-terminal part of the protein is unfolded and flexible³⁵. This region contains two conserved regions, CR1 and CR2. The conserved Leu-X-Cys-X-Glu (LXCXE) motive in CR2 is necessary for pRb-binding and the CR2 domain shares sequence similarity and transforming activity with proteins of other DNA tumour viruses as well (e.g. Adenovirus E1A, simian virus 40 large tumour antigen)³⁶. The C-terminal end contains a third conserved region, CR3, with two Cys-X-X-Cys (CXXC) motifs separated by 29 or 30 residues, necessary for the formation of a zinc binding fold and involved in protein stabilization and dimerization although dimerization has not yet been shown in vivo³⁵.

Figure 7. Alignment of HPV16, 18, 6 and 11 E7 amino acid sequences.

The conserved regions (CR) one to three are indicated with respect to the HPV16 E7 amino acid sequence³⁷.

The long pre-invasive phase, together with the fact the cervix is easily accessible for sampling and treatment, makes cervical cancer ideal for screening. Large parts of the world are covered by screening programs and this has led to a dramatic reduction in cervical cancer incidence with over 70 % reduction in some populations⁴ and women diagnosed by screening have a significantly higher chance of being cured than those diagnosed by detectable symptoms³⁸. However, many women are not yet covered by such programs and due to limitations of the available tests, a not insignificant portion of the screened women still gets cancer.

Screening prevention programs normally include primary screening, triage of equivocal results and colposcopy-guided biopsy in patients identified with abnormal screening results, followed by treatment and follow-up. Detection methods, start-points and screening intervals, as well as when and how to treat vary between settings. For primary screening, most programs use cervical cytology. In the widely used Pap test (Pap smear), first described by George Papanicolau in the 1940s, the morphology of the exfoliated cervical cells is examined under magnification³⁹. The sensitivity of the test is limited (around 50 % for CIN2/3) and the test needs to be repeated frequently, typically every 2-3 years, to be efficient²⁶. The evaluation identifies a spectrum of abnormalities and is subjective. Studies have shown that even experienced cytologists differ in their evaluation⁴⁰.

A refinement of the Pap smear, where the cervical cells are spread directly onto a microscope slide, is Liquid-based Cytology (LBC). In LBC the exfoliated cells are first transferred into a fixative liquid and then used to prepare monolayer slides. LBC has been shown to increase neither sensitivity nor specificity compared to the conventional pap smear⁴¹, however this technique has two major advantages; it creates more homogenously spread cell preparations and it leaves residual material for further analysis such as HPV testing, without requiring an extra sample. Furthermore, LBC samples can be processed using automated systems.

Two HPV vaccines are approved today, Cervarix (GlaxoSmithKline) and Gardasil (Merck). Both vaccines protect against HPV16 and 18 and Gardasil also targets HPV6 and 11 that cause genital warts. The vaccines are HPV L1 Virus-like-particle (VLP) vaccines. The vaccines have shown almost 100 %

protection against the targeted HPV types and offer cross-protection against closely related high-risk HPV types as well⁴². Vaccines for protection against additional carcinogenic types are under evaluation.

Since most women are infected at first intercourse, women should optimally be vaccinated before sexual debut. The vaccines have no therapeutic effect and vaccination of older, already exposed women is not protective⁴². For these generations of unvaccinated women screening will still be necessary. It is also important that vaccinated women continue to undergo regular cervical screening since the vaccines do not cover all high-risk types. There will also always be women who are not, for different reasons, covered by the vaccination program and need regular examinations.

Histological evaluation has long been the gold standard for diagnosis of cervical malignancy and is used to verify abnormal cytology results and solve inconclusive cytology. In most screening programs, the threshold for treatment is histological CIN2 or worse, despite the fact that 40 % of the CIN2 lesions regress spontaneously within two years⁴³. Most guidelines recommend treatment of the entire transformation zone at confirmed high- grade dysplasia since the entire zone is at risk. Fertility sparing methods, such as cryotherapy, cone-shaped excision and loop electrosurgical excision procedure (LEEP) are to be preferred, but in case of cancer, surgery and even radical hysterectomy, in combination with radiotherapy and chemotherapy, might be the best choice¹³. Most protocols recommend treating CIN2, rather than waiting and observing, due to the risk of missing underlying cancers^{44, 45}. This leads to a lot of overtreatment of lesions that would otherwise regress and which is extra problematic in women in childbearing age, due to the increased risk of preterm delivery following treatment⁴⁶.

The highest grade of CIN identified in a biopsy sample decides the management of the patient and since one patent can have several lesions of different severity and caused by different HPV types, the site of sampling becomes crucial^{10, 47}. This, in combination with large inter-observer varia- tions in the histological evaluation⁴⁰, contributes to the complexity of diagnosing cervical pre-cancer histologically. Multiple biopsies and immunohistochemical staining have been suggested to increase sensitivity and specificity for malignancy¹⁹.

A particular challenge in cervical cancer prevention is the management of low-grade and mildly atypical cervical abnormalities. The most common abnormality in cytology is Atypical Squamous Cells of Undetermined

Significance (ASC-US)⁴⁸. Millions of women are actually diagnosed with minor or equivocal abnormalities every year⁴⁹ and though most of these lesions will regress approximately 5% have underlying high-grade disease (CIN3+) that needs to be identified and treated⁵⁰. The high proportion of inconclusive cytology results, poor reproducibility in grading of lesions and large number of recalled and over-treated women highlights the need for alternative, preferably more objective approaches, such as biomarkers to solve some of the problems with today’s screening methods.

To increase the interobserver reproducibility, reduce subjectivity and/or increase specificity/sensitivity several biomarkers have been suggested in the diagnosis of cervical malignancy. Safer diagnosis has the potential of reducing the overtreatment of lesions that would otherwise regress, decrease the number of missed cases and hopefully make screening available for a larger part of the world.

HPV DNA testing is now widely used for triage of equivocal cytology findings and as co-testing to cytology, but has also been for suggested as a more sensitive test for high-grade dysplasia compared to cytology in primary screening⁵¹. The primary benefits of HPV DNA testing are the high sensitivity and the high negative predictive value and consequently, the screening interval after a negative HPV DNA test can safely be lengthened to six years due to the low risk of high-grade dysplasia⁵². However, a single DNA tests cannot discriminate transient infections from persistent and due to the high prevalence of transient HPV infections in the population, especially in young women, the utility of HPV testing in primary screening is somewhat limited. Another approach claimed to be more specific for malignancy than DNA testing is to measure HPV mRNA. mRNA testing is generally more specific and less sensitive than DNA at detecting high-grade dysplasia, depending the detection method used⁵³.

Women with ASC-US and borderline cytology should always be further managed in order to identify those women at risk due to the risk of under- lying high-grade disease (10-15 %)^{54, 55}. High-risk HPV DNA and mRNA tests normally perform better in ASC-US populations than borderline^{49, 56}. Since most low-grade lesions are HPV positive HPV testing is not selective for high-grade dysplasia in this group and especially not in women under 50⁴⁸. Thus, most guidelines recommend HPV DNA testing (in case of LBC), repetition of the smear and/or colposcopy for triage of ASC-US, while today only cytology and colposcopy is recommended for triage of borderline

samples⁵⁷. Other complementary tests are therefore needed for the large group of borderline diagnoses in cytology.

Some cellular markers have been suggested for more specific detection of cervical neoplasia. So far, the tumour suppressor gene p16 (p16^ink4a) is the most widely evaluated and promising cellular marker. p16 is overexpressed in high-grade dysplasia and cervical cancer due to E7-induced pRb degradation and loss of negative feedback⁵⁸. p16 cytology has been shown to have higher specificity but lower sensitivity for CIN2+ than HPV DNA testing⁵⁹. To further improve the test performance and facilitate the interpretation of the staining results, the combination of p16 and the proliferation marker Ki-67 has been introduced (mtm laboratories)⁶⁰. This dual staining protocol, judging a sample as positive based on the detection of at least one double stained cell irrespective of cellular morphology, has been shown to be more specific for detection of CIN2+ in the ASC-US and low- grade lesions compared to both HPV DNA testing and p16 single-stain cytology⁶¹.

Although the importance of the viral proteins for the malignant transformation is well established, no HPV protein test has yet been FDA approved, which is probably due to the lack of sensitive reagents. A few groups have managed to detect viral protein markers in clinical material e.g.

E7^{31, 62}, E6⁶³ and L1⁶⁴, but larger studies are needed to prove the clinical utility of these reagents.

The overall objective of this thesis was to develop immunological reagents against biomarkers associated with malignant transformation of cervical epithelium for improved diagnosis of cervical cancer.

The specific aims were to

Generate useful monoclonal antibodies against HPV E7

oncoproteins. The antibodies should have high affinity to oncogenic HPV types without cross-reactivity to cellular proteins.

Develop sensitive protocols for Immunocytochemistry,

Immunohistochemistry and other immunoassays to assist in early detection of cervical cancer.

Evaluate the diagnostic potential of the established methods using representative clinical specimens.

Three cervical cancer cell lines were obtained from the American Type Culture Collection (ATCC) and used as positive and negative controls throughout the studies; CaSki (ATCC: CRL-1550) expressing the HPV16 E7 protein, HeLa (ATCC: CCL-2) expressing the HPV18 E7 protein and C-33A (ATCC: HTB-31) negative for HPV DNA. The cell lines have the major advantage of being monoclonal, compared to the mixed nature of patient samples and are therefore trustable controls.

Two different immunization strategies were employed in this study; Site- specific immunization with peptides displayed on phage particles and immunization with full-length recombinant HPV E7 protein. Female Balb/c mice (B&K Universal) were used and kept at the Experimental Biomedicine facility, Sahlgrenska Academy at the University of Gothenburg. All experiments were approved by the Ethical Committee for Animal experimentation in Gothenburg, Sweden.

The mice were intraperitoneally injected with either 5 x 10¹¹ phage particles displaying E7 peptides, or first 25 μg and then 10 μg recombinant HPV16 or 18 E7 protein, in Sigma Adjuvant System. The serum reactivity against the full-length E7 protein was studied by ELISA and spleen cells from high-titer mice were collected for the establishment of antibody-producing hybridomas.

Newly isolated B-lymphocytes were fused with the mouse myeloma cell line P3X63Ag8.653 (ATCC: CRL-1580), by treating the cells with Polyethylene glycol (PEG) in a strictly controlled manner⁶⁵. The newly fused cells were grown in 96-well plates and selected through the addition of hypoxanthine, aminopterin and thymine (HAT) to the growth media (Fig. 8). The resulting hybridomas were then selected for reactivity to full-length HPV16 or 18 E7 in ELISA.

Figure 8. Fusion and in vitro selection of in HAT medium. When aminopterin (A) is added to the growth media normal DNA synthesis is inhibited and the cell has to use a salvage pathway (PW) which requires hypoxanthine (H) and thymine (T) for DNA synthesis. The mutant mouse myeloma cell line (M) is defective in one of the genes necessary for the salvage pathway and will die in the presence of aminopterin.

Myeloma cells fused with B-cells (B), however, are provided with the necessary genes and become immortal antibody-producing hybridomas. Non-fused B-lymphocytes only survive for up to two weeks in culture. Ig: Immunoglobulin.

Different immunological methods, primary ELISA, but also Western blot and Immunocytochemistry, were used to identify antibodies with the desired properties. Antibodies with reactivity to the low-risk HPV types were discarded and the high-risk specific ones were further evaluated. Promising hybridomas were expanded and cloned by limiting dilution. Antibody (Ab) isotypes were determined by ELISA using isotype-specific antibodies.

The mAbs used in all studies, were produced by in vitro cultivation in Dulbecco’s Modified Eagle Medium (DMEM, Sigma-Aldrich) with 2.5 % FBS and 1 % DMEM supplement in roller bottles. The antibodies were purified on Protein A and eluted at pH 4. Purity and concentration were determined by gel chromatography and by measuring the absorbance at 280 nm.

Phage display, display of peptides on the surface of the bacteriophage, has become a widely used technique to study protein interactions and produce large amounts of peptides attached to the phage carrier. Wild-type filamentous phage particles are rod-shaped and composed of a single- stranded phage genome, covered by approximately 2700 copies of the major pVIII protein and five copies each of the minor coat proteins pIII and pVI. By cloning a chosen nucleotide sequence in fusion with one of the phage coat proteins, either directly into a modified phage genome or into a separate phagemid, phage particles displaying the corresponding peptide on its surface are created⁶⁶. Phagemids are plasmids carrying a modified gene of one of the phage coat proteins together with antibiotic resistance and both a plasmid and a phage replication origin. A wild-type so called helper phage is then needed to support phage production. The resulting phage preparation is be a mixture of phage particles, carrying either the helper phage genome or the phagemid, all displaying the peptide encoded by the phagemid on a portion of its coat proteins. In this study, phage display was used for the construction of overlapping E7 peptide libraries for epitope determination, immunization of mice and mimotope mapping with commercial random peptide libraries.

Phage particles are immunogenic and can thus be used as immunogenic carriers of peptides⁶⁷. In paper I, phage display was used for Site-specific immunization, to direct the immune system to a specific pre-defined region of the E7 protein. A high-risk specific region of the HPV16 E7 sequence, corresponding to aa 33-60 of the E7 protein, was identified and cloned in phage vector f 88-4 , kindly provided by professor G.P. Smith (University of Missouri, Columbia, US), in fusion with the major coat protein pVIII. Phage particles were produced in E. coli, purified by double PEG precipitation and used to immunize mice. Spleen cells from mice with high HPV16 E7 titer were used to establish antibody-producing hybridomas, as described above.

Serum reactivity as well as the reactivity of the resulting antibodies was measured by ELISA against full-length E7 protein.

Since phage particles carry the sequence encoding the peptide it displays it is suitable for the construction of peptide libraries. Phage clones binding to a specific target, such as a mAb can be selected from the library in a panning

process. The sequences of the identified phage clones reveal the region, epitope or mimotope involved in binding.

In paper I and II, overlapping HPV E7 peptides, 16 to 37 aa long, were displayed on phage and antibody binding to the different clones was studied by ELISA. The antibodies to be examined were immobilized in microtiter wells and different phage clones were added each wells. After washing off the non-binding phage particles, phage binding was detected by incubation with a rabbit-anti-M13 antibody (raised against M13 phage particles at Fujirebio Diagnostics AB), a Horseradish peroxidase (HRP)-conjugated swine-anti-rabbit Immunoglobulin (Ig, Dako) and the HRP substrate Enhance K-blue (Neogen). The binding pattern to the different phage clones thereby directly revealed the E7 regions involved in binding.

In paper I and III, a commercially available random 7-mer peptide library displayed on phage (New England Biolabs) was used for mimotope identification. The library was panned against the mAbs in four rounds and binding clones were identified and sequenced. The consensus aa sequences were compared with the E7 aa sequence and aa essential for binding, the mimotope, were identified.

In paper III, three vectors for the display of peptides on either bacteriophage, in fusion with Glutathione-S-Transferase (GST), or in fusion with the cholera toxin B subunit (CTB) were generated. The vectors and inserts were designed so that the synthesized DNA could be cloned into any of the three vectors, depending on application. By arranging the peptide-encoding fragments, interspersed with linkers carrying restrictions sites for cloning, on the same DNA construct cloning of the entire library could easily be performed in a single reaction mix (Fig. 9). Here, the vector system was used to create libraries of HPV16 E7 peptides, 12 aa long, covering the entire E7 open reading-frame (ORF) and overlapping by 8 aa.

Figure 9. Arrangement of overlapping sequences for cloning into expression vectors.

Alternating inserts were inverted to minimize the number of bases to synthesize and shuffled to reduce secondary structure of overlapping sequences. The restriction sites (represented by X and Y) on the inserts and expression vectors were placed so that the vectors could be used interchangeably. The digested inserts were

dephosphorylated to prevent them from ligating to each other.

Four HPV16 E7 mAbs were epitope mapped by colony blot, using the E7 peptide library fused to GST. 135 random clones from the library were grown as single colonies in a grid pattern on LB plates and then transferred to sterile nitrocellulose filters by placing them on top of the colonies. The filters were then placed on fresh LB agar plates supplemented by IPTG to induce the expression of the GST fusions. The cells on the filters were lysed and the filters were incubated with the HPV16 E7 mAbs. MAb-binding was detected using HRP-conjugated goat anti-mouse Ig and visualized with HRP substrate 2-chloronaphtol. Plasmids of the positive clones on the filter were prepared from the same position on the corresponding LB-plate and sequenced.

The same overlapping HPV16 E7 peptides, displayed on phage particles, were used to map dominant B-cell epitopes in crude serum from mice immunized with full-length HPV16 E7 protein. The serum to be mapped was immobilized in microtiter wells. A mixture of 10⁸ phage particles, corresponding to nearly 5 x 10⁷ copies of each phage clone, was added to each well and 20 binding clones were sequenced after one round of panning.

The peptide sequences were compared to the HPV16 E7 sequence to identify linear B-cell epitopes.

ELISA (Enzyme-linked immunosorbent assay) offers the opportunity to study antibody interaction to the native antigen in solution. In this study, ELISA was used for different applications such as specificity studies with recombinant antigen, epitope mapping with phage particles and to monitor immune response in immunized mice. Although variables such as incubation times and Ab concentrations varied between applications, the key steps are common to all.

The catching antibody was immobilized in Maxisorp wells (Nunc A/S), either directly or, for unpurified Ab caught by an immobilized polyclonal goat-anti-mouse Ig antibody (Jackson Immunoresearch Laboratories). After a wash step, the antigen, diluted in blocking buffer was added. Antigen binding was detected using antibodies directed directly against the E7 protein or against a protein tag such as the phage particle or GST and an HRP- conjugated secondary antibody. Two different HRP substrates were used, either the o-Phenylenediamine (OPD) substrate (Sigma-Aldrich) or the more sensitive Enhance K-Blue (Neogen) substrate. Absorbance was measured at 450 or 620 nm in a microplate spectrophotometer (Vmax, Molecular Corporation), depending on the substrate used. An automatic plate washer, in strip mode with overflow, with 5mM Tris-HCl pH 7.8, 150 mM NaCl and 0.005 % Tween20. All incubations were at room temperature, either in a humid chamber or on a microplate shaker (900-1100 oscillations/min). When measuring serum reactivity against the E7 protein, GST-tagged E7 protein was directly coated in glutathione-coated wells (Pierce). Serial dilutions of serum were added to the E7-coated wells and anti-E7 reactivity was detected with HRP-conjugated rabbit anti-mouse (Dako).

The specificity of antibodies was also studied by Western blot. A panel of high- and low-risk (HPV1, 6, 11, 16, 18, 31, 33, 35 and 45) E7 proteins were separated by SDS-PAGE (NuPAGE) and blotted to PVDF membranes. The membranes were incubated with the E7 antibodies in 5 % non-fat dry milk and binding was detected with HRP-conjugated rabbit anti-mouse Ig (Dako) and chemiluminescent detection (ECL+, GE Healthcare).

Western blot was also used to ensure that the antibodies did not cross-react with other proteins present in cervical cells. Cell lysates were prepared from cervical cancer cell lines CaSki (expressing HPV16 E7 protein), HeLa (expressing HPV18 E7 protein) and C-33A (negative for HPV DNA). The cells were lysed by freeze-thawing and cell debris was removed by centrifugation. The total protein concentration was determined with the Bio- Rad protein assay kit and 50 μg of protein was loaded to each lane. The analysis was then done as described above.

All high-risk HPV-specific antibodies were evaluated using Immuno- cytochemistry (ICC). The newly harvested cervical cancer cell lines were either directly mounted on polysine slides and fixed in 10 % neutral buffered formalin or first fixed in the LBC preservative fluid (Thinprep Preservcyt, Hologic or Surepath, BD) and then mounted on slides. Different methods and buffers for antigen retrieval were evaluated such as heat-induced antigen retrieval in citrate or Tris buffer, to try to find the optimal protocol for each antibody. Other parameters, such as antibody concentrations, incubation times and incubation temperatures were also optimized in parallel.

The optimized protocols with the anti-HPV E7 mAbs E716-41 and E718-79 were, as presented in paper IV, evaluated on 49 patient cytology samples collected in Thinprep LBC fluid (Hologic). All samples were HPV DNA genotyped. 19 samples were judged as normal in the cytological evaluation.

The 20 dysplastic samples, all HPV16 or 18 positive, were from a referral population and cytology samples and biopsies for histological evaluation were taken at the same visit. All women provided informed consent.

The samples were either mRNA genotyped or evaluated in p16/Ki-67 cytology (CINtech PLUS, mtm laboratories AG). All mRNA genotype results were in agreement with the DNA genotyping. For p16/Ki-67 evaluation only double stained cells, with red nucleus and brown cytoplasm, were considered as positive. E7 cytology was interpreted as positive if cells abnormal cells were stained. All slides were evaluated by trained cytologists.

The cytologists knew the distribution of the specimens, but were blinded to the specific cytology diagnosis and HPV genotype result. The E7 cytology results were then correlated to HPV mRNA status, p16/Ki-67 cytology and histology results.

Using two different immunization strategies, over 100 hybridomas producing anti-HPV16 or HPV18 E7 antibodies were established. From these, high-risk HPV-specific antibodies with different properties were selected, with focus on finding those that detected the antigen by ICC, especially in LBC samples.

The identified antibodies had different specificities and properties depending on the immunogen and selection method used. (Fig. 10)

Figure 10. Schematic presentation of the workflow. Two different immunization strategies were used; Site-specific immunization with a high-risk specific HPV16 E7 peptide displayed on phage particles and immunization of recombinant full length E7 protein. Spleen cells from mice with high serum titers against full-length recombinant E7 were fused with a mouse myeloma cell line to get antibody-producing immortal hybridomas. The primary screening of the hybridomas was done by ELISA against full-length HPV16 and 18 E7 proteins. The reactivity of the selected hybridomas were further evaluated against a panel of high- and low risk HPV types in ELISA and Western blot and the epitopes were roughly mapped using overlapping E7 peptides displayed on phage. High-risk HPV-specific antibodies (Ab) were cloned and carefully evaluated against endogenous E7 in Immunocytochemistry (ICC) and Western blot. Mimotope analysis was performed for a subset of the high-risk Abs.

The HPV E7 protein is an antigen with more and less conserved regions.

Antibodies, especially those reacting with the highly conserved regions, will most probably detect the E7 protein of several HPV types. Whether type- specific antibodies or antibodies with broader specificity is preferred, will depend on the clinical application. Throughout these studies, high-risk HPV- specific antibodies were saved while the ones cross-reacting with the common, but not carcinogenic HPV-types were discarded. The reason for this was to avoid antibodies giving false positive results. This was done by immunizing with high-risk specific regions of the protein (site-specific immunization) and by selecting for high-risk specificity.

Figure 11. Phylogenetic tree based on the E7 amino acid sequences of 12 high-risk HPV types, two low-risk HPV and the probably carcinogenic HPV type 68. Due to the high sequence similarity between types, E7 antibodies often detect several E7 proteins, especially the more closely related such as HPV16, 31 and 35 or HPV18 and 45. The types in bold are the eight most common types, together found in approximately 90 % of all cervical cancers. The sequence differences between high- and low-risk HPV E7 made the establishment of high-risk specific E7 antibodies possible. The phylogenetic analysis is based on the multiple alignments of the E7 amino acid sequences in Clustal W. The evolutionary analyses were conducted in MEGA5.

HPV18 HPV45 HPV59 HPV39 HPV68 HPV51 HPV56 HPV16 HPV35 HPV31 HPV52 HPV33 HPV58 HPV11

HPV6 Low-risk

Phage display is an effective method to produce large amounts of peptides attached to an immunogenic carrier. Phage particles displaying pathogen- derived peptides induce protective immunity to that specific pathogen^{68, 69} and phage clones displaying mimotopes mimicking specific antibody epitopes and selected from random libraries, can activate immune responses against the original antigen^{70, 71}. Site-specific immunization with phage particles is an attractive approach when antibodies to a specific part of a protein are wanted for example in sandwich pairing, or to avoid highly immunogenic regions. Further, as phage particles are immunogenic and persist in the circulation or longer than free peptides that are rapidly cleared, site-specific immunization is a preferable alternative when antibodies to less immunogenic regions are needed.

In paper I, the immunogenicity of the phage particles was used to direct the immune response to a specific region of the HPV16 E7 oncoprotein thereby raising antibodies recognizing a pre-defined region of the endogenous E7 protein. A high-risk specific region, aa 33-60 of the HPV16 E7 protein was identified by sequence analysis. This fragment of HPV16 E7 has large sequence homology to the E7 protein of the closely related HPV types 31, 33 and 35, but not to the low-risk types or less related high-risk types, such as HPV18. The region was further chosen not to include the conserved regions, CR1-3. The corresponding nucleotide sequence was cloned into phage display vector f 88-4 , resulting mosaic phage particles with approximately 5 % of the pVIII proteins fused with the E7 peptide.

After seven immunizations with the phage preparations during the period of ten month, two out of three mice had high serum reactivity towards full- length HPV16 E7 and were sacrificed for hybridoma production. The fusion procedure resulted in three hybridomas, with reactivity to the HPV16 E7 protein, which were cloned by limiting dilution and evaluated using different immunological methods, in both paper I and II.

The three antibodies, named E716-1, E716-2 and E716-9 recognized full- length recombinant E7 of the three closely related high-risk types HPV16, 31 and 35, but not HPV 1, 6, 11, 18, 33 or 45, in ELISA and Western blot. The reactivity to the denatured protein in Western blot indicated that the antibodies recognize linear epitopes. The antibodies further detected the endogenous HPV16 E7 protein in lysates of the HPV16-positive cervical cancer cell line CaSki in Western blot, without cross-reactivity to other cellular proteins present in the lysates or to HPV18 E7 in HPV18-positive

HeLa cells. The mAbs specifically stained the HPV16 E7 protein in formalin fixed CaSki cells in ICC, indicating that the antibodies can be used to detect HPV16-induced malignancy in formalin fixed clinical samples.

Since the mice were immunized with an HPV16 E7 peptide corresponding to aa 33-60 of the HPV16 E7 fragment, the resulting antibodies recognized this region. To further characterize the epitopes fragment phage display was used.

Three peptides together covering the peptide used for immunization were displayed on phage particles. The reactivity of the antibodies to the different fragments was studied in ELISA and from the binding pattern, the epitope was narrowed down further. E716-1 and E716-9 recognized the two fragments aa 28-47 and 37-53, mapping the epitope to the overlapping region aa 37-47. Antibody E716-2 only bound one fragment aa 37-53, mapping the epitope of E716-2 to the junction between the other two fragments (aa 28-47 and 47-62) (Fig. 12).

To further confirm the mapping data and support the specificity results, mimope analysis was done with a commercially available random peptide library displayed on phage. The peptides of the library used here was only 7 aa long and therefore ideal for mapping of linear B-cell epitopes, though conformation dependent epitopes cannot be identified with such short peptides. The phage library, displaying the random peptides in fusion with the pIII coat protein, was panned against the antibodies in four rounds. Nine binding clones were sequenced for each mAb and the consensus sequence revealed the aa residues necessary for antigen binding, the mimotope. E716-1 and E716-9 had the exact same mimotope, Q A/S Q/R PD, corresponding to aa 44-48 on the HPV16 E7 sequence. This sequence is conserved between the three closely related types HPV16, 31 and 35 in agreement with the specificity study with recombinant protein and cannot be found in the other HPV E7 proteins. The mimotope of E716-2 was one position longer, Q A/S R PDR, corresponding to aa 44-49. This sequence can only be found in HPV16 E7, but since E716-2 also detects HPV31 and 35, the last position of the mimotope, the Arginine (-R-) is not crucial for binding. Altogether, the study shows how different methods can be used to complement and confirm results in the establishment of highly specific and well characterized antibodies for clinical applications.

Figure 12. Mapping results for the three HPV16 E7 antibodies raised by site- specific immunization. The antibodies were raised against amino acid (aa) 33-60 of the HPV16 E7 protein (underlined sequence). This region was chosen due to the sequence homology with a group of high-risk viruses (e.g. 31 and 35). The resulting antibodies detected mimotopes corresponding to aa 44-48 and 44-49 and recognize recombinant HPV16, 31 and 35 E7 but not HPV18 E7 or low-risk HPV6 or 11 E7 protein.

aa 37-53

aa 47-62 aa 28-47

Mimotope of E716-1 and E716-9 QA / S R / Q - P D Mimotope of E716-2 Q^{A / S R} - P D R

HPV 16 E E - E - D E I D G P A G Q A E - P D R A - - - H Y N I V T F C C K HPV 31 D E - E - D V I D S P A G Q A E - P D T S - - - N Y N I V T F C C Q HPV 35 E E - E E D T I D G P A G Q A K - P D T S - - - N Y N I V T S C C K HPV 33 D E - D E - G L D R P D G Q A Q - P A T A - - - D Y Y I V T C C H T HPV 58 D E - D E I G L D G P D G Q A Q - P A T A - - - N Y Y I V T C C Y T HPV 52 D E E D T D G V D R P D G Q A E - Q A T S - - - N Y Y I V T Y C H S HPV 18 E E - - N D E I D G V N H Q H L P A R R A E P Q R H T M L C M C C K

HPV 6 E - - - - D E V D E V D G Q D S Q P L K - - - Q H Y Q I V T C C C G HPV 11 E - - - - D E V D K V D K Q D A Q P L T - - - Q H Y Q I L T C C C G

0 1 2 3 4

E716-1 E716-2 E716-9

Abs 620 nm

MAb

A more widely used strategy for the establishment of monoclonal antibodies is to immunize the animals with full-length protein. In paper II, recombinant HPV16 and 18 E7 was used for immunization. The primary hybridoma selection was against recombinant HPV16 or 18 E7 proteins in ELISA and only hybridomas detecting either of the proteins were saved. This resulted in 57 hybridomas. The specificity was further evaluated in ELISA against a panel of high- and low-risk E7 proteins and those that cross-reacted with low- risk types were not further evaluated, resulting in 35 high-risk HPV-specific hybridomas. The majority of the antibodies recognized the E7 protein from more than one HPV type, which is not at all surprising, given the large sequence similarity throughout the viral genome. Most of the high-risk specific antibodies recognized a few, closely related HPV types, such as HPV18 and 45 or HPV16, 31 and 35.

The same panel of recombinant E7 proteins was used to study antibody specificity in Western blot. All the anti-E7 antibodies, selected for high-risk specificity against recombinant E7 in ELISA and functionality in ICC, had linear epitopes and recognized the E7 protein using Western blot. Overall, the specificity results from the Western blot analysis agreed with the ELISA analysis. The minor discrepancy seen between the two techniques is probably due to differences in affinities and how the specific epitopes are exposed.

As in paper I, overlapping peptides displayed on phage particles were used for epitope determination (Fig. 13). However, since the antibodies in paper II were raised against full-length protein fragments covering the entire HPV16 and HPV18 E7 ORFs were used. The peptides were of different length, from 16 to 37 aa long and displayed on phage particles in fusion with either coat protein pVIII or pIII. Antibody binding to the different peptides was then studied in ELISA to map the antibody epitopes. Epitope mapping with long peptides is a fast and straightforward technique to roughly identify the epitopes of a large number of antibodies targeting the same antigen. A separate set of clones has to be constructed for each antigen, but when this has been done, a large number of antibodies can easily be mapped in parallel.

The mapping data can be used to for example find possible ELISA pairs, predict antibody specificities or identify immunogenic regions on an antigen.

Figure 13. Epitope mapping of 57 antibodies (Ab) raised against the full-length HPV16 or 18 E7 protein. In total, 48 hybridomas recognized to the first 22 amino acids of the E7 protein, a region that includes conserved region (CR) 1 and several well conserved positions. The majority of these N-terminal antibodies recognized several HPV-types and only 14 were HPV16 or 18-specific. All antibodies that detected more central part of the protein were high-risk specific.

As expected, the 57 hybridomas targeted different regions of the E7 antigens.

However, the majority of the antibodies were directed to the N-terminal of the E7 protein and all 57 antibodies actually detected epitopes within aa 1-53, a region which is known to lack secondary structure³⁵. This is in agreement with earlier studies identifying the N-terminal half of the E7 protein as more immunogenic⁷². All 22 hybridomas that cross-reacted with the low-risk HPV types, detected aa 1 to 22, including CR1 which is highly conserved throughout all HPV E7 proteins. The majority of the N-terminal antibodies